A: Hey Phil!

Thanks for reaching out – that's a fantastic idea for an automated newsletter, and yes, AgenticFlow is absolutely designed to help you build something like this, even if you're new to workflows and coding! Our goal is to make this kind of powerful automation accessible.

Let's break down how you can achieve your automated newsletter:

1. Scraping YouTube/Other Sources for News (Hourly):
-
YouTube: You can use our YouTube Data API MCP (https://agenticflow.ai/mcp/youtube_data_api) to search for new videos on a specific topic or channel.

Other Web Sources (News sites, Blogs):
- If they have an RSS feed, our RSS MCP (https://agenticflow.ai/mcp/rss) is the easiest way to get new content.
- If no RSS, you can use our Web Scraping node or, for more complex sites, a tool like Firecrawl via its MCP (https://agenticflow.ai/mcp/firecrawl) to fetch new articles.

Scheduling: To run this hourly, you'd currently trigger your AgenticFlow workflow via its API endpoint using an external scheduler (many free/cheap options exist like EasyCron or even a simple script). A native built-in scheduler available.

2. Summarizing Topic & Writing Complete Newsletter:

Yes! The scraped content/video transcripts (you might need a speech-to-text step for YouTube videos if transcripts aren't available) would be fed into an LLM node.

You can use our built-in economical models (covered by your credits) or connect your Straico or Brevo (formerly Sendinblue, if they offer an LLM via API) API keys (BYOK).

Prompting the LLM: You'd instruct the LLM: "Take this collection of news snippets/summaries about [Your Topic]. Synthesize them into a coherent and engaging newsletter. Start with an overall summary, then detail a few key stories."

3. Generating an Image for the Topic:

Yes! Add a generate_image node. You can use keywords from your topic or a summary from the newsletter draft as a prompt for the image. This node typically uses services via our Pixel ML OpenRouter (separate pay-as-you-go credits) or your BYOK for an image generation API (like DALL-E via OpenAI, or if Straico offers image gen via API).

4. Manual Check/Approval:

Current Method: The workflow can send you an email (via Gmail MCP: https://agenticflow.ai/mcp/gmail) with the draft newsletter and image. You'd review it.

To continue automation after approval, you might reply to that email with "APPROVE" to trigger a second workflow (using email parsing), or click a link in the email that goes to a simple form whose submission triggers the next part.

Roadmap: A more integrated "Human Approval Step" node is something we're considering (vote here: https://agenticflow.featurebase.app/).

5. Automatically Post to Social Media Channels:

Yes! Once approved:
- Use our LinkedIn MCP (https://agenticflow.ai/mcp/linkedin) to post.
- Use our Facebook Pages MCP (https://agenticflow.ai/mcp/facebook_pages) to post.

For other platforms, if they have an API, use the API Call node.

6. Connecting Straico/Brevo:

- Straico: If Straico offers a REST API for its LLM or image capabilities, you'd connect it using our API Call node within your workflow.
- Brevo (Sendinblue): We have an MCP for Brevo (https://agenticflow.ai/mcp/brevo). You'd use this primarily for sending the email newsletter, rather than AI content generation.

Can We Help You Build It?
Absolutely! Especially if you're on Tier 3 or 4 (which gives you priority support), we're happy to help guide you.

Start Simple: Break it down. First, try to get the news aggregation working. Then focus on the summarization, then image, then posting.

Use Our Resources: Check our YouTube channel (we're adding more tutorials!) and documentation.

Community & Support: Ask specific questions in our Discord (https://discord.gg/WtZxB5jB) or email support@agenticflow.ai.

Office Hours / 1:1: For a more complex build like this, joining our weekly office hours or requesting a brief 1:1 session (especially for T3/4) can be very effective for getting personalized guidance.

You don't need to be a coder to do this with AgenticFlow. It will take some learning to connect the pieces, but the visual builder and natural language prompting make it much more accessible. We're here to support you!

Best regards,
Sean